Individual assets and problem behaviors in at-risk adolescents: A longitudinal cross-lagged analysis

Abstract

Autoregressive cross-lagged structural equation modeling was conducted to examine longitudinal relationships between individual assets (social competence, positive values and identity) and problem behaviors in 373 adolescents (174 boys, 199 girls) who participated in a prospective study on the effects of prenatal cocaine exposure from birth. More behavioral problems at age 12 were related to fewer individual assets at age 15, while greater individual assets were related to more behavioral problems, with a non-significant yet nuanced (p = .076) gender difference. More problem behaviors were associated with decreased individual assets in girls, yet greater individual assets were associated with more problem behaviors in boys. Efforts to promote individual assets may not lower the risk of engaging in problem behaviors especially among boys. Continued studies into adulthood will uncover how individual assets and problem behaviors in childhood and adolescence may affect social and vocational adjustment in this high risk population.

Keywords: Adolescents; Cross-lagged; Individual assets; Problem behavior; Structural equation modeling.

Figure 1. Autoregressive cross-lagged structural equation model on the association between individual assets and problem behaviors in at-risk adolescents for the overall study sample (N = 373)

χ² (222) = 399.92, p < .001, CFI = .950, TLI = .932, RMSEA = .046 (.039 - .054). Single-arrowed lines represent standardized path coefficients whereas double-arrowed lines represent correlations. Solid lines indicate statistically significant coefficients (p < .05) whereas dotted lines indicate non-significant coefficients (p > .05). Estimates are adjusted for prenatal exposure to cocaine, gender, race, violence exposure, HOME scores, and IQ. Covariates and errors associated with the measurement model were not reported for parsimony. Learning = commitment to learning; value = positive value; competence = social competence; identity = positive identity; rule-break = rule-breaking behavior. *p < .05; ** p < .01; *** p <.001.

Figure 2. Standardized coefficients from the multiple-group cross-lagged structural equation model, with female data (n = 199) presented before the slash and male's (n = 174) after the slash

χ²(429) = 635.85, p < .001, CFI = .944, TLI = .926, RMSEA = .036 (.030 - .042). Single-arrowed lines represent path coefficients whereas double-arrowed lines represent correlations. Solid lines indicate statistically significant coefficients (p < .05) for at least one of groups whereas dotted lines indicate nonsignificant coefficients (p > .05). Estimates are adjusted for prenatal exposure to cocaine, gender, race, violence exposure, HOME scores, and IQ. Covariates and errors associated with the measurement model were not reported for parsimony. Learning = commitment to learning; value = positive value; competence = social competence; identity = positive identity; rule-break = rule-breaking behavior. *p < .05; ** p < .01; *** p <.001; ^ p < .10.